Now that mddev_suspend() doean't rely on 'mddev->pers' to be set, it's
safe to remove such checking.
This will also allow the array to be suspended even before the array
is ran.
Signed-off-by: Yu Kuai <yukuai3@huawei.com>
Signed-off-by: Song Liu <song@kernel.org>
Link: https://lore.kernel.org/r/20230825030956.1527023-8-yukuai1@huaweicloud.com
Now that mddev_suspend() doean't rely on 'mddev->pers' to be set, it's
safe to remove such checking.
This will also allow the array to be suspended even before the array
is ran.
Signed-off-by: Yu Kuai <yukuai3@huawei.com>
Signed-off-by: Song Liu <song@kernel.org>
Link: https://lore.kernel.org/r/20230825030956.1527023-7-yukuai1@huaweicloud.com
'active_io' used to be initialized while the array is running, and
'mddev->pers' is set while the array is running as well. Hence caller
must hold 'reconfig_mutex' and guarantee 'mddev->pers' is set before
calling mddev_suspend().
Now that 'active_io' is initialized when mddev is allocated, such
restriction doesn't exist anymore. In the meantime, follow up patches
will refactor mddev_suspend(), hence add checking for 'mddev->pers' to
prevent null-ptr-deref.
Signed-off-by: Yu Kuai <yukuai3@huawei.com>
Signed-off-by: Song Liu <song@kernel.org>
Link: https://lore.kernel.org/r/20230825030956.1527023-4-yukuai1@huaweicloud.com
Currently 'writes_pending' is initialized in pers->run for raid1/5/10,
and it's freed while deleing mddev, instead of pers->free. pers->run can
be called multiple times before mddev is deleted, and a helper
mddev_init_writes_pending() is used to prevent 'writes_pending' to be
initialized multiple times, this usage is safe but a litter weird.
On the other hand, 'writes_pending' is only initialized for raid1/5/10,
however, it's used in common layer, for example:
array_state_store
set_in_sync
if (!mddev->in_sync) -> in_sync is used for all levels
// access writes_pending
There might be some implicit dependency that I don't recognized to make
sure 'writes_pending' can only be accessed for raid1/5/10, but there are
no comments about that.
By the way, it make sense to initialize 'writes_pending' in common layer
because there are already three levels use it.
Signed-off-by: Yu Kuai <yukuai3@huawei.com>
Signed-off-by: Song Liu <song@kernel.org>
Link: https://lore.kernel.org/r/20230825030956.1527023-3-yukuai1@huaweicloud.com
'active_io' is used for mddev_suspend() and it's initialized in
md_run(), this restrict that 'reconfig_mutex' must be held and
"mddev->pers" must be set before calling mddev_suspend().
Initialize 'active_io' early so that mddev_suspend() is safe to call
once mddev is allocated, this will be helpful to refactor
mddev_suspend() in following patches.
Signed-off-by: Yu Kuai <yukuai3@huawei.com>
Signed-off-by: Song Liu <song@kernel.org>
Link: https://lore.kernel.org/r/20230825030956.1527023-2-yukuai1@huaweicloud.com
Before this patch, for read-only array:
md_check_recovery() check that 'MD_RECOVERY_NEEDED' is set, then it will
call remove_and_add_spares() directly to try to remove and add rdevs
from array.
After this patch:
1) md_check_recovery() check that 'MD_RECOVERY_NEEDED' is set, and the
worker 'sync_work' is not pending, and there are rdevs can be added
or removed, then it will queue new work md_start_sync();
2) md_start_sync() will call remove_and_add_spares() and exist;
This change make sure that array reconfiguration is independent from
daemon, and it'll be much easier to synchronize it with io, consier
that io may rely on daemon thread to be done.
Also fix a problem that 'pers->spars_active' is called after
remove_and_add_spares(), which order is wrong, because spares must
active first, and then remove_and_add_spares() can add spares to the
array, like what read-write case does:
1) daemon set 'MD_RECOVERY_RUNNING', register new sync thread to do
recovery;
2) recovery is done, md_do_sync() set 'MD_RECOVERY_DONE' before return;
3) daemon call 'pers->spars_active', and clear 'MD_RECOVERY_RUNNING';
4) in the next round of daemon, call remove_and_add_spares() to add
spares to the array.
Signed-off-by: Yu Kuai <yukuai3@huawei.com>
Signed-off-by: Song Liu <song@kernel.org>
Link: https://lore.kernel.org/r/20230825031622.1530464-8-yukuai1@huaweicloud.com
There are no functional changes, just to make the code simpler and
prepare to delay remove_and_add_spares() to md_start_sync().
Signed-off-by: Yu Kuai <yukuai3@huawei.com>
Reviewed-by: Xiao Ni <xni@redhat.com>
Signed-off-by: Song Liu <song@kernel.org>
Link: https://lore.kernel.org/r/20230825031622.1530464-6-yukuai1@huaweicloud.com
Before this patch, for read-write array:
1) md_check_recover() found that something need to be done, and it'll
try to grab 'reconfig_mutex'. The case that md_check_recover() need
to do something:
- array is not suspend;
- super_block need to be updated;
- 'MD_RECOVERY_NEEDED' or 'MD_RECOVERY_DONE' is set;
- unusual case related to safemode;
2) if 'MD_RECOVERY_RUNNING' is not set, and 'MD_RECOVERY_NEEDED' is set,
md_check_recover() will try to choose a sync action, and then queue a
work md_start_sync().
3) md_start_sync() register sync_thread;
After this patch,
1) is the same;
2) if 'MD_RECOVERY_RUNNING' is not set, and 'MD_RECOVERY_NEEDED' is set,
queue a work md_start_sync() directly;
3) md_start_sync() will try to choose a sync action, and then register
sync_thread();
Because 'MD_RECOVERY_RUNNING' is cleared when sync_thread is done, 2)
and 3) and md_do_sync() is always ran in serial and they can never
concurrent, this change should not introduce any behavior change for now.
Also fix a problem that md_start_sync() can clear 'MD_RECOVERY_RUNNING'
without protection in error path, which might affect the logical in
md_check_recovery().
The advantage to change this is that array reconfiguration is
independent from daemon now, and it'll be much easier to synchronize it
with io, consider that io may rely on daemon thread to be done.
Signed-off-by: Yu Kuai <yukuai3@huawei.com>
Reviewed-by: Xiao Ni <xni@redhat.com>
Signed-off-by: Song Liu <song@kernel.org>
Link: https://lore.kernel.org/r/20230825031622.1530464-4-yukuai1@huaweicloud.com
There are no functional changes, on the one hand make the code cleaner,
on the other hand prevent following checkpatch error in the next patch to
delay choosing sync action to md_start_sync().
ERROR: do not use assignment in if condition
+ } else if ((spares = remove_and_add_spares(mddev, NULL))) {
Signed-off-by: Yu Kuai <yukuai3@huawei.com>
Reviewed-by: Xiao Ni <xni@redhat.com>
Signed-off-by: Song Liu <song@kernel.org>
Link: https://lore.kernel.org/r/20230825031622.1530464-3-yukuai1@huaweicloud.com
It's a little weird to borrow 'del_work' for md_start_sync(), declare
a new work_struct 'sync_work' for md_start_sync().
Signed-off-by: Yu Kuai <yukuai3@huawei.com>
Reviewed-by: Xiao Ni <xni@redhat.com>
Signed-off-by: Song Liu <song@kernel.org>
Link: https://lore.kernel.org/r/20230825031622.1530464-2-yukuai1@huaweicloud.com
If there are multiple arrays in system and one mddevice is marked
with MD_DELETED and md_seq_next() is called in the middle of removal
then it _get()s proper device but it may _put() deleted one. As a result,
active counter may never be zeroed for mddevice and it cannot
be removed.
Put the device which has been _get with previous md_seq_next() call.
Cc: stable@vger.kernel.org
Fixes: 12a6caf273 ("md: only delete entries from all_mddevs when the disk is freed")
Reported-by: AceLan Kao <acelan@gmail.com>
Closes: https://bugzilla.kernel.org/show_bug.cgi?id=217798
Cc: Yu Kuai <yukuai3@huawei.com>
Signed-off-by: Mariusz Tkaczyk <mariusz.tkaczyk@linux.intel.com>
Signed-off-by: Song Liu <song@kernel.org>
Link: https://lore.kernel.org/r/20230914152416.10819-1-mariusz.tkaczyk@linux.intel.com
Commit a1d7671910 ("md: use mddev->external to select holder in
export_rdev()") fix the problem that 'claim_rdev' is used for
blkdev_get_by_dev() while 'rdev' is used for blkdev_put().
However, if mddev->external is changed from 0 to 1, then 'rdev' is used
for blkdev_get_by_dev() while 'claim_rdev' is used for blkdev_put(). And
this problem can be reporduced reliably by following:
New file: mdadm/tests/23rdev-lifetime
devname=${dev0##*/}
devt=`cat /sys/block/$devname/dev`
pid=""
runtime=2
clean_up_test() {
pill -9 $pid
echo clear > /sys/block/md0/md/array_state
}
trap 'clean_up_test' EXIT
add_by_sysfs() {
while true; do
echo $devt > /sys/block/md0/md/new_dev
done
}
remove_by_sysfs(){
while true; do
echo remove > /sys/block/md0/md/dev-${devname}/state
done
}
echo md0 > /sys/module/md_mod/parameters/new_array || die "create md0 failed"
add_by_sysfs &
pid="$pid $!"
remove_by_sysfs &
pid="$pid $!"
sleep $runtime
exit 0
Test cmd:
./test --save-logs --logdir=/tmp/ --keep-going --dev=loop --tests=23rdev-lifetime
Test result:
------------[ cut here ]------------
WARNING: CPU: 0 PID: 960 at block/bdev.c:618 blkdev_put+0x27c/0x330
Modules linked in: multipath md_mod loop
CPU: 0 PID: 960 Comm: test Not tainted 6.5.0-rc2-00121-g01e55c376936-dirty #50
RIP: 0010:blkdev_put+0x27c/0x330
Call Trace:
<TASK>
export_rdev.isra.23+0x50/0xa0 [md_mod]
mddev_unlock+0x19d/0x300 [md_mod]
rdev_attr_store+0xec/0x190 [md_mod]
sysfs_kf_write+0x52/0x70
kernfs_fop_write_iter+0x19a/0x2a0
vfs_write+0x3b5/0x770
ksys_write+0x74/0x150
__x64_sys_write+0x22/0x30
do_syscall_64+0x40/0x90
entry_SYSCALL_64_after_hwframe+0x63/0xcd
Fix the problem by recording if 'rdev' is used as holder.
Fixes: a1d7671910 ("md: use mddev->external to select holder in export_rdev()")
Signed-off-by: Yu Kuai <yukuai3@huawei.com>
Signed-off-by: Song Liu <song@kernel.org>
Link: https://lore.kernel.org/r/20230825025532.1523008-3-yukuai1@huaweicloud.com
Except for initial reference, mddev->kobject is referenced by
rdev->kobject, and if the last rdev is freed, there is no guarantee that
mddev is still valid. Hence mddev should not be used anymore after
export_rdev().
This problem can be triggered by following test for mdadm at very
low rate:
New file: mdadm/tests/23rdev-lifetime
devname=${dev0##*/}
devt=`cat /sys/block/$devname/dev`
pid=""
runtime=2
clean_up_test() {
pill -9 $pid
echo clear > /sys/block/md0/md/array_state
}
trap 'clean_up_test' EXIT
add_by_sysfs() {
while true; do
echo $devt > /sys/block/md0/md/new_dev
done
}
remove_by_sysfs(){
while true; do
echo remove > /sys/block/md0/md/dev-${devname}/state
done
}
echo md0 > /sys/module/md_mod/parameters/new_array || die "create md0 failed"
add_by_sysfs &
pid="$pid $!"
remove_by_sysfs &
pid="$pid $!"
sleep $runtime
exit 0
Test cmd:
./test --save-logs --logdir=/tmp/ --keep-going --dev=loop --tests=23rdev-lifetime
Test result:
general protection fault, probably for non-canonical address 0x6b6b6b6b6b6b6bcb: 0000 [#4] PREEMPT SMP
CPU: 0 PID: 1292 Comm: test Tainted: G D W 6.5.0-rc2-00121-g01e55c376936 #562
RIP: 0010:md_wakeup_thread+0x9e/0x320 [md_mod]
Call Trace:
<TASK>
mddev_unlock+0x1b6/0x310 [md_mod]
rdev_attr_store+0xec/0x190 [md_mod]
sysfs_kf_write+0x52/0x70
kernfs_fop_write_iter+0x19a/0x2a0
vfs_write+0x3b5/0x770
ksys_write+0x74/0x150
__x64_sys_write+0x22/0x30
do_syscall_64+0x40/0x90
entry_SYSCALL_64_after_hwframe+0x63/0xcd
Fix this problem by don't dereference mddev after export_rdev().
Fixes: 3ce94ce5d0 ("md: fix duplicate filename for rdev")
Signed-off-by: Yu Kuai <yukuai3@huawei.com>
Signed-off-by: Song Liu <song@kernel.org>
Link: https://lore.kernel.org/r/20230825025532.1523008-2-yukuai1@huaweicloud.com
-----BEGIN PGP SIGNATURE-----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=p1bd
-----END PGP SIGNATURE-----
Merge tag 'for-6.6/block-2023-08-28' of git://git.kernel.dk/linux
Pull block updates from Jens Axboe:
"Pretty quiet round for this release. This contains:
- Add support for zoned storage to ublk (Andreas, Ming)
- Series improving performance for drivers that mark themselves as
needing a blocking context for issue (Bart)
- Cleanup the flush logic (Chengming)
- sed opal keyring support (Greg)
- Fixes and improvements to the integrity support (Jinyoung)
- Add some exports for bcachefs that we can hopefully delete again in
the future (Kent)
- deadline throttling fix (Zhiguo)
- Series allowing building the kernel without buffer_head support
(Christoph)
- Sanitize the bio page adding flow (Christoph)
- Write back cache fixes (Christoph)
- MD updates via Song:
- Fix perf regression for raid0 large sequential writes (Jan)
- Fix split bio iostat for raid0 (David)
- Various raid1 fixes (Heinz, Xueshi)
- raid6test build fixes (WANG)
- Deprecate bitmap file support (Christoph)
- Fix deadlock with md sync thread (Yu)
- Refactor md io accounting (Yu)
- Various non-urgent fixes (Li, Yu, Jack)
- Various fixes and cleanups (Arnd, Azeem, Chengming, Damien, Li,
Ming, Nitesh, Ruan, Tejun, Thomas, Xu)"
* tag 'for-6.6/block-2023-08-28' of git://git.kernel.dk/linux: (113 commits)
block: use strscpy() to instead of strncpy()
block: sed-opal: keyring support for SED keys
block: sed-opal: Implement IOC_OPAL_REVERT_LSP
block: sed-opal: Implement IOC_OPAL_DISCOVERY
blk-mq: prealloc tags when increase tagset nr_hw_queues
blk-mq: delete redundant tagset map update when fallback
blk-mq: fix tags leak when shrink nr_hw_queues
ublk: zoned: support REQ_OP_ZONE_RESET_ALL
md: raid0: account for split bio in iostat accounting
md/raid0: Fix performance regression for large sequential writes
md/raid0: Factor out helper for mapping and submitting a bio
md raid1: allow writebehind to work on any leg device set WriteMostly
md/raid1: hold the barrier until handle_read_error() finishes
md/raid1: free the r1bio before waiting for blocked rdev
md/raid1: call free_r1bio() before allow_barrier() in raid_end_bio_io()
blk-cgroup: Fix NULL deref caused by blkg_policy_data being installed before init
drivers/rnbd: restore sysfs interface to rnbd-client
md/raid5-cache: fix null-ptr-deref for r5l_flush_stripe_to_raid()
raid6: test: only check for Altivec if building on powerpc hosts
raid6: test: make sure all intermediate and artifact files are .gitignored
...
Commit ba9d9f1a707f ("Revert "md: unlock mddev before reap sync_thread in
action_store"") removed the scenario of calling md_unregister_thread()
without holding mddev->reconfig_mutex, so add a lock holding check before
acquiring mddev->sync_thread by passing mdev to md_unregister_thread().
Signed-off-by: Li Lingfeng <lilingfeng3@huawei.com>
Reviewed-by: Yu Kuai <yukuai3@huawei.com>
Link: https://lore.kernel.org/r/20230803071711.2546560-1-lilingfeng@huaweicloud.com
Signed-off-by: Song Liu <song@kernel.org>
memalloc_noio_save() is called for the first mddev_suspend(), and
repeated mddev_suspend() only increase 'suspended'. However,
memalloc_noio_restore() is also called for the first mddev_resume(),
which means that memory reclaim will be enabled before the last
mddev_resume() is called, while the array is still suspended.
Fix this problem by restore 'noio_flag' for the last mddev_resume().
Fixes: 78f57ef9d5 ("md: use memalloc scope APIs in mddev_suspend()/mddev_resume()")
Signed-off-by: Yu Kuai <yukuai3@huawei.com>
Link: https://lore.kernel.org/r/20230628012931.88911-3-yukuai1@huaweicloud.com
Signed-off-by: Song Liu <song@kernel.org>
Some levels doesn't implement "pers->quiesce", for example
raid0_quiesce() is empty, and now that all levels will drop 'active_io'
until io is done, wait for 'active_io' to be 0 is enough to make sure all
normal io is done, and percpu_ref_kill() for 'active_io' will make sure
no new normal io can be dispatched. There is no need to call
"pers->quiesce" anymore from mddev_suspend().
Signed-off-by: Yu Kuai <yukuai3@huawei.com>
Link: https://lore.kernel.org/r/20230628012931.88911-2-yukuai1@huaweicloud.com
Signed-off-by: Song Liu <song@kernel.org>
Currently, 'active_io' is grabbed before make_reqeust() is called, and
it's dropped immediately make_reqeust() returns. Hence 'active_io'
actually means io is dispatching, not io is inflight.
For raid0 and raid456 that io accounting is enabled, 'active_io' will
also be grabbed when bio is cloned for io accounting, and this 'active_io'
is dropped until io is done.
Always clone new bio so that 'active_io' will mean that io is inflight,
raid1 and raid10 will switch to use this method in later patches.
Now that bio will be cloned even if io accounting is disabled, also
rename related structure from '*_acct_*' to '*_clone_*'.
Signed-off-by: Yu Kuai <yukuai3@huawei.com>
Reviewed-by: Xiao Ni <xni@redhat.com>
Signed-off-by: Song Liu <song@kernel.org>
Link: https://lore.kernel.org/r/20230621165110.1498313-3-yukuai1@huaweicloud.com
'io_acct_set' is only used for raid0 and raid456, prepare to use it for
raid1 and raid10, so that io accounting from different levels can be
consistent.
By the way, follow up patches will also use this io clone mechanism to
make sure 'active_io' represents in flight io, not io that is dispatching,
so that mddev_suspend will wait for io to be done as designed.
Signed-off-by: Yu Kuai <yukuai3@huawei.com>
Reviewed-by: Xiao Ni <xni@redhat.com>
Signed-off-by: Song Liu <song@kernel.org>
Link: https://lore.kernel.org/r/20230621165110.1498313-2-yukuai1@huaweicloud.com
The support for bitmaps on files is a very bad idea abusing various kernel
APIs, and fundamentally requires the file to not be on the actual array
without a way to check that this is actually the case. Add a deprecation
warning to see if we might be able to eventually drop it.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Himanshu Madhani <himanshu.madhani@oracle.com>
Signed-off-by: Song Liu <song@kernel.org>
Link: https://lore.kernel.org/r/20230615064840.629492-12-hch@lst.de
The support for write intent bitmaps in files on an external files in md
is a hot mess that abuses ->bmap to map file offsets into physical device
objects, and also abuses buffer_heads in a creative way.
Make this code optional so that MD can be built into future kernels
without buffer_head support, and so that we can eventually deprecate it.
Note this does not affect the internal bitmap support, which has none of
the problems.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Himanshu Madhani <himanshu.madhani@oracle.com>
Signed-off-by: Song Liu <song@kernel.org>
Link: https://lore.kernel.org/r/20230615064840.629492-11-hch@lst.de
For md_check_recovery():
1) if 'MD_RECOVERY_RUNING' is not set, register new sync_thread.
2) if 'MD_RECOVERY_RUNING' is set:
a) if 'MD_RECOVERY_DONE' is not set, don't do anything, wait for
md_do_sync() to be done.
b) if 'MD_RECOVERY_DONE' is set, unregister sync_thread. Current code
expects that sync_thread is not NULL, otherwise new sync_thread will
be registered, which will corrupt the array.
Make sure md_check_recovery() won't register new sync_thread if
'MD_RECOVERY_RUNING' is still set, and a new WARN_ON_ONCE() is added for
the above corruption,
Signed-off-by: Yu Kuai <yukuai3@huawei.com>
Reviewed-by: Xiao Ni <xni@redhat.com>
Signed-off-by: Song Liu <song@kernel.org>
Link: https://lore.kernel.org/r/20230529132037.2124527-7-yukuai1@huaweicloud.com
md_reap_sync_thread() is just replaced with wait_event(resync_wait, ...)
from action_store(), just make sure action_store() will still wait for
everything to be done in md_reap_sync_thread().
Signed-off-by: Yu Kuai <yukuai3@huawei.com>
Reviewd-by: Xiao Ni <xni@redhat.com>
Signed-off-by: Song Liu <song@kernel.org>
Link: https://lore.kernel.org/r/20230529132037.2124527-6-yukuai1@huaweicloud.com
Our test found a following deadlock in raid10:
1) Issue a normal write, and such write failed:
raid10_end_write_request
set_bit(R10BIO_WriteError, &r10_bio->state)
one_write_done
reschedule_retry
// later from md thread
raid10d
handle_write_completed
list_add(&r10_bio->retry_list, &conf->bio_end_io_list)
// later from md thread
raid10d
if (!test_bit(MD_SB_CHANGE_PENDING, &mddev->sb_flags))
list_move(conf->bio_end_io_list.prev, &tmp)
r10_bio = list_first_entry(&tmp, struct r10bio, retry_list)
raid_end_bio_io(r10_bio)
Dependency chain 1: normal io is waiting for updating superblock
2) Trigger a recovery:
raid10_sync_request
raise_barrier
Dependency chain 2: sync thread is waiting for normal io
3) echo idle/frozen to sync_action:
action_store
mddev_lock
md_unregister_thread
kthread_stop
Dependency chain 3: drop 'reconfig_mutex' is waiting for sync thread
4) md thread can't update superblock:
raid10d
md_check_recovery
if (mddev_trylock(mddev))
md_update_sb
Dependency chain 4: update superblock is waiting for 'reconfig_mutex'
Hence cyclic dependency exist, in order to fix the problem, we must
break one of them. Dependency 1 and 2 can't be broken because they are
foundation design. Dependency 4 may be possible if it can be guaranteed
that no io can be inflight, however, this requires a new mechanism which
seems complex. Dependency 3 is a good choice, because idle/frozen only
requires sync thread to finish, which can be done asynchronously that is
already implemented, and 'reconfig_mutex' is not needed anymore.
This patch switch 'idle' and 'frozen' to wait sync thread to be done
asynchronously, and this patch also add a sequence counter to record how
many times sync thread is done, so that 'idle' won't keep waiting on new
started sync thread.
Noted that raid456 has similiar deadlock([1]), and it's verified[2] this
deadlock can be fixed by this patch as well.
[1] https://lore.kernel.org/linux-raid/5ed54ffc-ce82-bf66-4eff-390cb23bc1ac@molgen.mpg.de/T/#t
[2] https://lore.kernel.org/linux-raid/e9067438-d713-f5f3-0d3d-9e6b0e9efa0e@huaweicloud.com/
Signed-off-by: Yu Kuai <yukuai3@huawei.com>
Signed-off-by: Song Liu <song@kernel.org>
Link: https://lore.kernel.org/r/20230529132037.2124527-5-yukuai1@huaweicloud.com
Currently, for idle and frozen, action_store will hold 'reconfig_mutex'
and call md_reap_sync_thread() to stop sync thread, however, this will
cause deadlock (explained in the next patch). In order to fix the
problem, following patch will release 'reconfig_mutex' and wait on
'resync_wait', like md_set_readonly() and do_md_stop() does.
Consider that action_store() will set/clear 'MD_RECOVERY_FROZEN'
unconditionally, which might cause unexpected problems, for example,
frozen just set 'MD_RECOVERY_FROZEN' and is still in progress, while
'idle' clear 'MD_RECOVERY_FROZEN' and new sync thread is started, which
might starve in progress frozen. A mutex is added to synchronize idle
and frozen from action_store().
Signed-off-by: Yu Kuai <yukuai3@huawei.com>
Signed-off-by: Song Liu <song@kernel.org>
Link: https://lore.kernel.org/r/20230529132037.2124527-4-yukuai1@huaweicloud.com
Prepare to handle 'idle' and 'frozen' differently to fix a deadlock, there
are no functional changes except that MD_RECOVERY_RUNNING is checked
again after 'reconfig_mutex' is held.
Signed-off-by: Yu Kuai <yukuai3@huawei.com>
Signed-off-by: Song Liu <song@kernel.org>
Link: https://lore.kernel.org/r/20230529132037.2124527-3-yukuai1@huaweicloud.com
This reverts commit 9dfbdafda3.
Because it will introduce a defect that sync_thread can be running while
MD_RECOVERY_RUNNING is cleared, which will cause some unexpected problems,
for example:
list_add corruption. prev->next should be next (ffff0001ac1daba0), but was ffff0000ce1a02a0. (prev=ffff0000ce1a02a0).
Call trace:
__list_add_valid+0xfc/0x140
insert_work+0x78/0x1a0
__queue_work+0x500/0xcf4
queue_work_on+0xe8/0x12c
md_check_recovery+0xa34/0xf30
raid10d+0xb8/0x900 [raid10]
md_thread+0x16c/0x2cc
kthread+0x1a4/0x1ec
ret_from_fork+0x10/0x18
This is because work is requeued while it's still inside workqueue:
t1: t2:
action_store
mddev_lock
if (mddev->sync_thread)
mddev_unlock
md_unregister_thread
// first sync_thread is done
md_check_recovery
mddev_try_lock
/*
* once MD_RECOVERY_DONE is set, new sync_thread
* can start.
*/
set_bit(MD_RECOVERY_RUNNING, &mddev->recovery)
INIT_WORK(&mddev->del_work, md_start_sync)
queue_work(md_misc_wq, &mddev->del_work)
test_and_set_bit(WORK_STRUCT_PENDING_BIT, ...)
// set pending bit
insert_work
list_add_tail
mddev_unlock
mddev_lock_nointr
md_reap_sync_thread
// MD_RECOVERY_RUNNING is cleared
mddev_unlock
t3:
// before queued work started from t2
md_check_recovery
// MD_RECOVERY_RUNNING is not set, a new sync_thread can be started
INIT_WORK(&mddev->del_work, md_start_sync)
work->data = 0
// work pending bit is cleared
queue_work(md_misc_wq, &mddev->del_work)
insert_work
list_add_tail
// list is corrupted
The above commit is reverted to fix the problem, the deadlock this
commit tries to fix will be fixed in following patches.
Signed-off-by: Yu Kuai <yukuai3@huawei.com>
Signed-off-by: Song Liu <song@kernel.org>
Link: https://lore.kernel.org/r/20230529132037.2124527-2-yukuai1@huaweicloud.com
__md_stop_writes() and __md_stop() will modify many fields that are
protected by 'reconfig_mutex', and all the callers will grab
'reconfig_mutex' except for md_stop().
Also, update md_stop() to make certain 'reconfig_mutex' is held using
lockdep_assert_held().
Fixes: 9d09e663d5 ("dm: raid456 basic support")
Signed-off-by: Yu Kuai <yukuai3@huawei.com>
Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Commit 3ce94ce5d0 ("md: fix duplicate filename for rdev") introduce a
new lock 'delete_mutex', and trigger a new deadlock:
t1: remove rdev t2: sysfs writer
rdev_attr_store rdev_attr_store
mddev_lock
state_store
md_kick_rdev_from_array
lock delete_mutex
list_add mddev->deleting
unlock delete_mutex
mddev_unlock
mddev_lock
...
lock delete_mutex
kobject_del
// wait for sysfs writers to be done
mddev_unlock
lock delete_mutex
// wait for delete_mutex, deadlock
'delete_mutex' is used to protect the list 'mddev->deleting', turns out
that this list can be protected by 'reconfig_mutex' directly, and this
lock can be removed.
Fix this problem by removing the lock, and use 'reconfig_mutex' to
protect the list. mddev_unlock() will move this list to a local list to
be handled after 'reconfig_mutex' is dropped.
Fixes: 3ce94ce5d0 ("md: fix duplicate filename for rdev")
Signed-off-by: Yu Kuai <yukuai3@huawei.com>
Signed-off-by: Song Liu <song@kernel.org>
Link: https://lore.kernel.org/r/20230621142933.1395629-1-yukuai1@huaweicloud.com
If bitmap is enabled, bitmap must update before submitting write io, this
is why unplug callback must move these io to 'conf->pending_io_list' if
'current->bio_list' is not empty, which will suffer performance
degradation.
A new helper md_bitmap_unplug_async() is introduced to submit bitmap io
in a kworker, so that submit bitmap io in raid10_unplug() doesn't require
that 'current->bio_list' is empty.
This patch prepare to limit the number of plugged bio.
Signed-off-by: Yu Kuai <yukuai3@huawei.com>
Signed-off-by: Song Liu <song@kernel.org>
Link: https://lore.kernel.org/r/20230529131106.2123367-6-yukuai1@huaweicloud.com
Commit 1a855a0606 ("md: fix bug with re-adding of partially recovered
device.") only add device which is set to In_sync. But it let devices
without metadata cannot be added when they should be.
Commit bf572541ab ("md: fix regression with re-adding devices to arrays
with no metadata") fix the above issue, it set device without metadata to
In_sync when add new disk.
However, after commit f466722ca6 ("md: Change handling of save_raid_disk
and metadata update during recovery.") deletes changes of the first patch,
setting In_sync for devcie without metadata is meanless because the flag
will be cleared soon and will not be used during this period. Clean it up.
Signed-off-by: Li Nan <linan122@huawei.com>
Reviewed-by: Yu Kuai <yukuai3@huawei.com>
Signed-off-by: Song Liu <song@kernel.org>
Link: https://lore.kernel.org/r/20230527101851.3266500-2-linan666@huaweicloud.com
Currently, there are many places that md_thread can be accessed without
protection, following are known scenarios that can cause
null-ptr-dereference or uaf:
1) sync_thread that is allocated and started from md_start_sync()
2) mddev->thread can be accessed directly from timeout_store() and
md_bitmap_daemon_work()
3) md_unregister_thread() from action_store().
Currently, a global spinlock 'pers_lock' is borrowed to protect
'mddev->thread' in some places, this problem can be fixed likewise,
however, use a global lock for all the cases is not good.
Fix this problem by protecting all md_thread with rcu.
Signed-off-by: Yu Kuai <yukuai3@huawei.com>
Signed-off-by: Song Liu <song@kernel.org>
Link: https://lore.kernel.org/r/20230523021017.3048783-6-yukuai1@huaweicloud.com
md_wakeup_thread() can't wakeup md_thread->tsk if md_thread->run is
still in progress, and in some cases md_thread->tsk need to be woke up
directly, like md_set_readonly() and do_md_stop().
Commit 9dfbdafda3 ("md: unlock mddev before reap sync_thread in
action_store") introduce a new scenario where unregister sync_thread is
not protected by 'reconfig_mutex', this can cause null-ptr-deference in
theroy:
t1: md_set_readonly t2: action_store
md_unregister_thread
// 'reconfig_mutex' is not held
// 'reconfig_mutex' is held by caller
if (mddev->sync_thread)
thread = *threadp
*threadp = NULL
wake_up_process(mddev->sync_thread->tsk)
// null-ptr-deference
Fix this problem by factoring out a helper to wake up md_thread directly,
so that 'sync_thread' won't be accessed multiple times from the reader
side. This helper also prepare to protect md_thread with rcu.
Noted that later patches is going to fix that unregister sync_thread is
not protected by 'reconfig_mutex' from action_store().
Signed-off-by: Yu Kuai <yukuai3@huawei.com>
Signed-off-by: Song Liu <song@kernel.org>
Link: https://lore.kernel.org/r/20230523021017.3048783-2-yukuai1@huaweicloud.com
Commit 5792a2856a ("[PATCH] md: avoid a deadlock when removing a device
from an md array via sysfs") delays the deletion of rdev, however, this
introduces a window that rdev can be added again while the deletion is
not done yet, and sysfs will complain about duplicate filename.
Follow up patches try to fix this problem by flushing workqueue, however,
flush_rdev_wq() is just dead code, the progress in
md_kick_rdev_from_array():
1) list_del_rcu(&rdev->same_set);
2) synchronize_rcu();
3) queue_work(md_rdev_misc_wq, &rdev->del_work);
So in flush_rdev_wq(), if rdev is found in the list, work_pending() can
never pass, in the meantime, if work is queued, then rdev can never be
found in the list.
flush_rdev_wq() can be replaced by flush_workqueue() directly, however,
this approach is not good:
- the workqueue is global, this synchronization for all raid disks is
not necessary.
- flush_workqueue can't be called under 'reconfig_mutex', there is still
a small window between flush_workqueue() and mddev_lock() that other
contexts can queue new work, hence the problem is not solved completely.
sysfs already has apis to support delete itself through writer, and
these apis, specifically sysfs_break/unbreak_active_protection(), is used
to support deleting rdev synchronously. Therefore, the above commit can be
reverted, and sysfs duplicate filename can be avoided.
A new mdadm regression test is proposed as well([1]).
[1] https://lore.kernel.org/linux-raid/20230428062845.1975462-1-yukuai1@huaweicloud.com/
Fixes: 5792a2856a ("[PATCH] md: avoid a deadlock when removing a device from an md array via sysfs")
Signed-off-by: Yu Kuai <yukuai3@huawei.com>
Signed-off-by: Song Liu <song@kernel.org>
Link: https://lore.kernel.org/r/20230523012727.3042247-1-yukuai1@huaweicloud.com
There is no input check when echo md/max_read_errors and overflow might
occur. Add check of input number.
Fixes: 1e50915fe0 ("raid: improve MD/raid10 handling of correctable read errors.")
Signed-off-by: Li Nan <linan122@huawei.com>
Reviewed-by: Yu Kuai <yukuai3@huawei.com>
Signed-off-by: Song Liu <song@kernel.org>
Link: https://lore.kernel.org/r/20230522072535.1523740-3-linan666@huaweicloud.com
There is no input check when echo md/safe_mode_delay in safe_delay_store().
And msec might also overflow when HZ < 1000 in safe_delay_show(), Fix it by
checking overflow in safe_delay_store() and use unsigned long conversion in
safe_delay_show().
Fixes: 72e02075a3 ("md: factor out parsing of fixed-point numbers")
Signed-off-by: Li Nan <linan122@huawei.com>
Signed-off-by: Song Liu <song@kernel.org>
Link: https://lore.kernel.org/r/20230522072535.1523740-2-linan666@huaweicloud.com
If reshape is in progress and io across reshape_position is issued, such
io will wait for reshape to make progress(see details in the case that
make_stripe_request() return STRIPE_SCHEDULE_AND_RETRY).
It has been reported several times that if system reboot while growing
raid5 to raid6, array assemble will hang infinitely([1, 2]). This is
because following deadlock is triggered:
1) a normal io is waiting for reshape to progress, this io can be from
system-udevd or mdadm.
2) while assemble, mdadm tries to suspend the array, hence
'reconfig_mutex' is held and mddev_suspend() must wait for normal io
to be done.
3) daemon thread can't start reshape because 'reconfig_mutex' can't be
held.
1) and 3) is unbreakable because they're foundation design. In order to
break 2), following is possible solutions that I can think of:
a) Let mddev_suspend() fail is not a good option, because this will
break many scenarios since mddev_suspend() doesn't fail before.
b) Fail the io that is waiting for reshape to make progress from
mddev_suspend().
c) Return false for the io that is waiting for reshape to make
progress from raid5_make_request(), and these io will wait for
suspend to be done in md_handle_request(), where 'active_io' is
not grabbed.
c) sounds better than b), however, b) is used because it's easy and
straightforward, and it's verified that mdadm can assemble in this case.
On the other hand, c) breaks the logic that mddev_suspend() will wait
for submitted io to be completely handled.
Fix the problem by checking reshape in mddev_suspend(), if reshape can't
make progress and there are still some io waiting for reshape, fail
those io.
[1] https://lore.kernel.org/all/CAFig2csUV2QiomUhj_t3dPOgV300dbQ6XtM9ygKPdXJFSH__Nw@mail.gmail.com/
[2] https://lore.kernel.org/all/CAO2ABipzbw6QL5eNa44CQHjiVa-LTvS696Mh9QaTw+qsUKFUCw@mail.gmail.com/
Reported-by: Jove <jovetoo@gmail.com>
Reported-by: David Gilmour <dgilmour76@gmail.com>
Signed-off-by: Yu Kuai <yukuai3@huawei.com>
Signed-off-by: Song Liu <song@kernel.org>
Link: https://lore.kernel.org/r/20230512015610.821290-6-yukuai1@huaweicloud.com
The only overlap between the block open flags mapped into the fmode_t and
other uses of fmode_t are FMODE_READ and FMODE_WRITE. Define a new
blk_mode_t instead for use in blkdev_get_by_{dev,path}, ->open and
->ioctl and stop abusing fmode_t.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Acked-by: Jack Wang <jinpu.wang@ionos.com> [rnbd]
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Christian Brauner <brauner@kernel.org>
Link: https://lore.kernel.org/r/20230608110258.189493-28-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
The current interface for exclusive opens is rather confusing as it
requires both the FMODE_EXCL flag and a holder. Remove the need to pass
FMODE_EXCL and just key off the exclusive open off a non-NULL holder.
For blkdev_put this requires adding the holder argument, which provides
better debug checking that only the holder actually releases the hold,
but at the same time allows removing the now superfluous mode argument.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Acked-by: Christian Brauner <brauner@kernel.org>
Acked-by: David Sterba <dsterba@suse.com> [btrfs]
Acked-by: Jack Wang <jinpu.wang@ionos.com> [rnbd]
Link: https://lore.kernel.org/r/20230608110258.189493-16-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
The mode argument to the ->release block_device_operation is never used,
so remove it.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Acked-by: Christian Brauner <brauner@kernel.org>
Acked-by: Jack Wang <jinpu.wang@ionos.com> [rnbd]
Link: https://lore.kernel.org/r/20230608110258.189493-10-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
->open is only called on the whole device. Make that explicit by
passing a gendisk instead of the block_device.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Acked-by: Christian Brauner <brauner@kernel.org>
Acked-by: Jack Wang <jinpu.wang@ionos.com> [rnbd]
Link: https://lore.kernel.org/r/20230608110258.189493-9-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
bdev_check_media_change should only ever be called for the whole device.
Pass a gendisk to make that explicit and rename the function to
disk_check_media_change.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Acked-by: Christian Brauner <brauner@kernel.org>
Link: https://lore.kernel.org/r/20230608110258.189493-8-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Add a new blk_holder_ops structure, which is passed to blkdev_get_by_* and
installed in the block_device for exclusive claims. It will be used to
allow the block layer to call back into the user of the block device for
thing like notification of a removed device or a device resize.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Jan Kara <jack@suse.cz>
Acked-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Link: https://lore.kernel.org/r/20230601094459.1350643-10-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
The md-raid superblock writing code uses bio_add_page() to add a page to a
newly created bio. bio_add_page() can fail, but the return value is never
checked.
Use __bio_add_page() as adding a single page to a newly created bio is
guaranteed to succeed.
This brings us a step closer to marking bio_add_page() as __must_check.
Signed-of_-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Reviewed-by: Damien Le Moal <damien.lemoal@opensource.wdc.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Acked-by: Song Liu <song@kernel.org>
Link: https://lore.kernel.org/r/ca196f5e650e318106dbb4496eb6cbac4bc800bd.1685532726.git.johannes.thumshirn@wdc.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
